Comment from Cong Chen
AnonymousOtherIndividual
Summary: The commenter provides technical feedback regarding the statistical allocation ratios for umbrella trials in drug development. They argue that the current recommendations may not be optimal for trials where experimental arms enter at different times or have overlapping populations, citing specific research for a more complex formula.
The recommended allocation ratio is optimal only when all the experimental arms enter into the study at the same time and the number of experimental arms is definite. However, in practice, both assumptions can be wrong for a bona fide umbrella trial and as a result the optimal allocation strategy may be different. For example, in an umbrella trial with two experimental arms, the optimal allocation ratio is 1+(√(3-p)-1)/p where p is the proportion of overlapping between the two experimental arms when both experimental are recruiting and is one otherwise. For more details, see Ren Y, Li N, Chen C. Statistical Considerations of Phase 3 Umbrella Trials Allowing Adding One Treatment Arm Mid-Trial. Contemporary Clinical Trials 2021. https://doi.org/10.1016/j.cct.2021.106538.